pythonlessons/mltu ? reverse-engineered prompt

Build me a small Python library that makes training machine learning models easier for both TensorFlow and PyTorch.

I want it to include the kind of helpers you’d need for real projects, like loading data, preparing datasets, training loops, saving models, and checking results. It should be useful for things like text recognition, OCR, captcha reading, handwriting recognition, speech to text, translation, and object detection with YOLOv8.

Please also add a set of simple tutorial examples that show how to use the library for a few different training tasks, so someone can learn by running the code and tweaking it. Keep the examples practical and easy to follow, with separate folders for each demo.

If you need to, look up current docs online for TensorFlow, PyTorch, and YOLOv8 so the examples use the right modern approach.